Improving multipath live streaming performance with Markov Decision Processes
Vinh Quang Bui, Weiping Zhu · 2007
Recent studies have demonstrated the benefit of multipath live contents streaming in the Internet best-effort environment. In this paper, we propose an approach to en- hance performance of the multipath streaming by using Markov Decision Processes. We formulate the multipath live streaming problem with the minimum distortion rate objective as a Markov Decision Process (MDP) and solve the formulated MDP on the fly using a computational effective algorithm namely OPI i.e. On- line Policy Iteration. We demonstrate the superior performance of the proposed approach against the classical schemes including Weighted Round Robin (WRR) and Join the Shortest Queue (JSQ) through intensive simulation using Ns-2. The approach proposed is transport protocol independent, which implies that it can be used friendly with both UDP and TCP streaming.